Regression is a popular machine learning technique used to predict target data based on a set of features. In classification, we train the system to assign distinctive labels to the object (Orange or Bananas in our previous example). Regression differs in the sense that here we are dealing with continuous variables.
Suppose that you have twenty thousand rupees with you and are planning to buy a new phone. You open up an e-commerce site and search for phones. A hundred results appear. You see names like Redmi, Oppo, Vivo, Realme, Asus, Nokia, and so on, all in your target range of ten thousand to twenty thousand. You are clearly confused. Therefore, you decide to have a deeper comparison.
You start plotting the prices of each phone and start noting down the features offered. As you watch closely, you realize that some features (like support for 4g and dual sim support) are offered by most of the phones while the presence or absence of some features affects the price. Models with additional features like full HD display and quad camera are priced above fifteen thousand while the others are at the lower end of the price range. Then you do a trade-off and decide on the phone that suits you best.
Now let us look at how we use similar plotting of continuous values to predict certain aspects of people. One of the favorite target groups for such regression analysis in our daily lives is, unfortunately, women. The character of a girl or a woman is often predicted by the time at which she goes out in the public.
Let us talk about our favorite random person, X. X is a college-going girl for now. If she is back home or hostel by six in the evening, she is supposed to be well mannered. If she stays out later than six or (god forbid) six-thirty, something is not right! For working women, we have benevolently pushed the boundaries to seven. Don’t even ask me about eight or later. When the brutal and horrific news of Nirbhaya came out several years ago, some people around me were wondering why she was out at that hour of the day. I could not agree with their line of thought at all. However, the prime convict Mukesh Singh had the same questions. Frankly, I do not think he is the sort of person you wanna hold up as your role model.
Timings are only one of the features employed to predict the character. Another common feature is the length of her skirt, with knees being a major benchmark. Another feature that holds interest is the length of your hair. It is interesting because this feature works in contrasting ways for men and women. The longer, the better for women. The shorter, the better for men.
How regressive, right?